Data set trained on 8 years of cat picture data from Poofy and Windmill, which in return creates an unlimited amount of computer generated cat postures mutating into each other through a latency step process trained on a style gan3 model.

This model pipeline from @latententities comes up with original versions of familiar cat identities that we would not come up with in a lifetime of explorations.

The export is then parented to a persian cat skeleton that uses a houdini kinefx setup by @blemoine that randomly scales the bones while maintaining the overall shape with additive animation.

The A.I. voice has been interpreted by @werkaanzee using a sub oscillator running through a bold analogue filter modulated by a classic sample and hold circuit to produce an endless variation of
tone. ✨

The three production approaches are contrasted together to create a new aesthetic & sound process @geoffreylillemon_.

Follow @poofy_x_windmill_a.i to see posts of cat data set images, generated outputs, and further exploration.

Portrait Outputs:

  
︎
︎